MAGL blockade alleviates steroid-induced femoral head osteonecrosis by reprogramming BMSC fate in rat

Abstract

The leading cause of steroid-induced femoral head osteonecrosis (ONFH) is the imbalance of bone homeostasis. Bone marrow-derived mesenchymal stem cell (BMSC) differentiation and fate are closely associated with bone homeostasis imbalance. Blocking monoacylglycerol lipase (MAGL) could effectively ameliorate ONFH by mitigating oxidative stress and apoptosis in BMSCs induced by glucocorticoids (GC). Nevertheless, whether MAGL inhibition can modulate the balance during BMSC differentiation, and therefore improve ONFH, remains elusive. Our study indicates that MAGL inhibition can effectively rescue the enhanced BMSC adipogenic differentiation caused by GC and promote their differentiation toward osteogenic lineages. Cannabinoid receptor 2 (CB2) is the direct downstream target of MAGL in BMSCs, rather than cannabinoid receptor 1(CB1). Using RNA sequencing analyses and a series of in vitro experiments, we confirm that the MAGL blockade-induced enhancement of BMSC osteogenic differentiation is primarily mediated by the phosphoinositide 3-kinases (PI3K)/ the serine/threonine kinase (AKT)/ (glycogen synthase kinase-3 beta) GSK3β pathway. Additionally, MAGL blockade can also reduce GC-induced bone resorption by directly suppressing osteoclastogenesis and indirectly reducing the expression of receptor activator of nuclear factor kappa-Β ligand (RANKL) in BMSCs. Thus, our study proposes that the therapeutic effect of MAGL blockade on ONFH is partly mediated by restoring the balance of bone homeostasis and MAGL may be an effective therapeutic target for ONFH.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00018-024-05443-5.

Introduction

High-dose administration of glucocorticoids (GCs) is now considered a major risk factor for osteonecrosis [1]. Numerous studies have shown that GCs have toxic effects on bone marrow stromal cells (BMSCs) and inhibit their osteogenic differentiation [2]. Concurrently, excess GCs promote osteoclast proliferation and differentiation [3]. Bone homeostasis imbalance caused by GCs induces osteoporosis and ultimately leads to osteonecrosis of the femoral head (ONFH). Thus, further research on the mechanisms of GC-induced bone homeostasis imbalance may help to discover novel femoral head-protective strategies.

Our previous study demonstrated that GCs exert a direct effect on BMSCs by inducing apoptosis, leading to reduced bone formation and strength [4]. In addition, the capability of GCs to change the differentiation direction of BMSCs is also an important factor in ONFH development. Studies have reported that dexamethasone (DEX) triggers adipogenic differentiation in pluripotent marrow cell lineages, such as D1 cells derived from BALB/c mice, resulting in the accumulation of intracellular lipid vesicles, upregulation of peroxisome proliferator-activated receptor-γ (PPARγ) gene expression, and downregulation of core-binding factor alpha1 (CBFA1)/RUNX2 gene expression [5]. DEX could also shift BMSCs from osteoblasts to adipocytes by CCAAT enhancer-binding protein (CEBP) promoter methylation [6]. High-dose GCs promote osteoclast proliferation by increasing the secretion of RANKL in osteoblasts and osteocytes and enhancing bone resorption [7]. On the other hand, the latest data indicate that RANKL expression from bone marrow adipocytes plays a key role in osteoclastogenesis [8, 9]. Considering that GCs promote BMSC adipocyte differentiation, which leads to increased RANKL levels and bone resorption enhancement, BMSC differentiation fate may be the main factor that affects GC-induced bone homeostasis imbalance. However, there is no direct evidence for this hypothesis.

Thus far, the mechanisms that control the direction of BMSC differentiation remain to be determined. One possible mediator of BMSC fate is the endocannabinoid system (EC) [4, 10]. As a central hub of 2-arachidonoyl glycerol (2AG) metabolism in ECs, MAGL is currently considered to be involved in multiple pathological processes, including neuroinflammation, cancer, metabolic disorders, and pain [11]. 2AG, the cannabinoid receptor ligand, binds to cannabinoid receptor 1 (CB1) and cannabinoid receptor 2 (CB2) with varying degrees of selectivity [12]. Thus, inhibition of MAGL can produce a medicinal effect by activating CB1 or CB2. Recent studies have demonstrated that mice lacking either CB1 or CB2 develop abnormal bone phenotypes [13]. AM251, a CB1 receptor antagonist, or CB1 knockout, can enhance adipogenic differentiation and suppress osteogenic differentiation of BMSCs [14]. In contrast, several CB2-selective agonists play opposing regulatory roles in BMSC differentiation [15]. Additionally, MAGL can also directly regulate osteoclast formation and bone loss. Our previous study showed that MAGL inhibition acted on BMSCs and reduced GC-induced oxidative stress and apoptosis [4]. However, the role of MAGL in the direction of BMSC differentiation and bone homeostasis imbalance in ONFH remains elusive.

In this study, we examined the potential of MAGL inhibition to confer protection against GC-induced ONFH by altering the direction of differentiation of BMSCs and investigated the underlying signaling pathways in a rat model of GC-induced ONFH. Our findings unequivocally demonstrated that MAGL blockade effectively ameliorates ONFH by promoting osteogenic differentiation and inhibiting adipogenic differentiation in BMSCs. According to pathway analysis, MAGL inhibition exerted its protective effects via CB2-dependent negative regulation of the PI3K/AKT/GSK3β pathway. Remarkably, MAGL blockade also significantly attenuated GC-induced bone resorption and restored bone homeostasis. Taken together, our results suggest that MAGL inhibition may represent a promising therapeutic strategy for GC-induced ONFH.

Materials and methods

Reagents and antibodies

Table S1-S4 provides a comprehensive description of the antibodies, primers, plasmids, siRNAs, chemicals and cell lines used in this study.

Cell culture

BMSCs were obtained from the femurs and tibias of 6-week-old SD rats following established procedures [16]. The cells were maintained in α-MEM medium (Cyagen, China) supplemented with penicillin, streptomycin, and fetal bovine serum (FBS) (Cyagen, China) at 37 °C. Cells in passages 3 to 6 were used for all experiments. The medium was refreshed every 3 days, and cells were passaged when they reached 80% confluence. A DMEM-based medium was used to culture RAW264.7 cells supplemented with 10% FBS, 1% penicillin/streptomycin, and 50 ng/mL RANKL.

Osteogenic induction

Forty-eight hours after seeding, BMSC osteogenic differentiation was induced. The osteogenic differentiation medium was purchased from Cyagen Biosciences (Soochow, China). The medium was exchanged every 3 days.

Adipogenic induction

Adipogenic differentiation medium was purchased from Cyagen Biosciences (Soochow, China). When the BMSCs had propagated to a density of 100%, the culture medium was switched to adipogenic differentiation medium A, and after 3 days of intramuscular culture, the medium was replaced with adipogenic differentiation medium B. The BMSCs were then maintained in medium B for 24 h. The two media were altered 3–5 times, and then the BMSCs were maintained with medium B, which was changed every 2–3 days until large lipid droplets were apparent.

CCK8 assay

We plated BMSCs (1 × 10³ cells/well) in 96-well plates and then cultured them in a medium containing a variety of chemical concentrations. Then, A fresh medium containing 10% CCK8 stock solution (Beyotime Biotech, China) was added to the cells and incubated at 37 °C. We measured the absorbance at 450 nm after two hours.

Western blotting

The protein isolation protocols were performed following our previous publication [4]. Briefly, the cell lysates were subjected to electrophoresis and transferred onto nitrocellulose membranes. The membranes were blocked using QuickBlock Buffer (Beyotime Biotech) for 1 h. Subsequently, the membranes were incubated overnight at 4 °C with primary antibodies. After washing thrice with PBS, the membranes were exposed to the relevant secondary antibodies and detected through chemiluminescence using Pierce ECL. The resulting autoradiograph was analyzed using Image Lab 3.0 software.

Quantitative real-time PCR

Total RNA was extracted using TRIzol reagent (Beyotime Biotech, China). The purity and concentration of RNA were assessed using a Nano Photometer spectrophotometer (Thermo Fisher, USA). Subsequently, cDNA was synthesized from the RNA using a cDNA kit (BioRad). Real-time PCR was performed using qPCR Master Mix (Biotium), and the mRNA expression levels were analyzed using Bio-Rad CFX Manager software with GAPDH as the internal control. The primers used were provided by Sangon Biotech (Shanghai, China).

shRNA, siRNA, and plasmid interference

The shRNA, siRNAs, and plasmids were obtained from GenePharma (Shanghai, China) and lentiviral transfection was performed according to the manufacturer’s instructions. The transfection protocol and interference were previously described in our published study [4]. Relative gene and protein expression levels were used to assess the efficiency of the transfection.

Cell immunofluorescence (IF) staining

The medium was first removed from the 24-well plates. After three PBS washes, the cells were fixed with paraformaldehyde for 15 min before being promptly bilayed with 0.2% Triton X-100 for one hour. The cells were then inhibited for 30 min using serum. The coverslips were then treated with primary antibodies overnight at 4 °C. Following rinsing with PBS, the coverslips were incubated with secondary antibodies for 2 h in the dark. The coverslips were rinsed again with PBS before being stained with DAPI. The total number of positively stained cells within a view area of a fluorescence microscope was counted to measure the amount of protein expression.

Alkaline phosphatase (ALP) staining

After being planted in 12-well plates with 2.5 × 10⁴ cells per well, BMSC was osteogenically induced for a week. The cells were subsequently fixed for 15 min in 4% paraformaldehyde. The wells were introduced to the BCIP/NBT working solution (300 µL) after three PBS washes. A one-hour incubation in the dark with the working solution was performed on the cells. Finally, using a laser microscope, the number of ALP-positive cells in each area was counted.

ALP activity assay

As directed by the manufacturer, the detection buffer, chromogenic substrate, and standard were fabricated. After lysing the cells from the various groups, the lysates were centrifuged at 3000 rpm for 15 min. For each sample well, blank well, and control well, the prepared reagents were added. A microplate reader was used to measure the OD value at 410 nm.

Alizarin red staining

2.5 ×10⁴ BMSCs were planted in each well of 12-well plates, and they were then treated for 2 weeks in an osteogenic differentiation medium. A 15-minute fixation in 95% alcohol was then performed on the cells. The wells were put into an Alizarin Red solution (500 µL) after three PBS washes. The working solution was incubated with the cells for 15 min at room temperature. Then, ddH₂O was three times circled each well to look for calcium nodules using a laser microscope. Finally, 10% cetylpyridinium chloride was used to dissolve the calcium nodules, and the OD value was measured at 570 nm.

Oil red O staining

BMSCs were seeded in 12-well plates at a density of 2.5 ×10⁴ cells per well and incubated for two weeks in adipogenic differentiation medium. The medium was removed, and the cells were washed twice with PBS before being fixed for 20 min in 4% paraformaldehyde. The cells were washed 3 times with ddH₂O before being incubated for five minutes in 60% isopropanol. The cells were then incubated for 10 min in an Oil Red O (Sigma, USA) solution. The cells were subsequently resuspended in 60% isopropanol and ddH₂O. The tissue staining procedure was identical to that for cultured cells. A laser microscope observed and captured images.

TRAP staining of cells

Five days were spent incubating RAW264.7 cells in an osteoclast differentiation medium. The medium was eliminated, and we washed the cells three times in PBS before fixing them for 20 min in 4% paraformaldehyde and personalizing them for 5 min in 0.3% Triton X-100. As described previously [17], a leukocyte acid phosphatase staining kit (Sigma, USA) was then used to detect osteoclasts. Osteoclasts are TRAP-positive cells with more than three nuclei.

RNA sequencing

BMSCs were cultured in 6-well plates. We divided the cells into two groups and performed three replications for each group. One group was MAGL knockout cells, and the other group was transduced with a negative control shRNA. The two groups of cells were treated with DEX 1 µM for 2 days. Then the TRIzol reagent (Beyotime) was used to extract the total RNA. The final subsequent sequencing analysis of the samples was performed by OE Biotech (Shanghai, China). 150-bp paired-end reads were generated from the libraries using an Illumina platform. An alignment of the clean reads to Rattus norvegicus. Rnor 6.0 was then conducted. Cufflinks were used to calculate FPKM for each gene, and HTSeq-count was used to calculate read counts. To perform differential expression analysis, we used the R package DESeq (2012). We determined significant differential expression by setting the P value at 0.05 and the fold change at 2 or the fold change at 0.5. Gene expression patterns in different samples and groups were analyzed using hierarchical cluster analysis (HCA). DEGs were enriched using the hypergeometric distribution in GO and KEGG pathway enrichment analyses.

Measurement by liquid chromatography/mass spectrometry (LC-MS) of intracellular 2AG levels

BMSCs (1 ×10⁵ cells/well) were cultured in 6-well plates and treated with MJN110 (0 µM, 1 µM, or 10 µM) in the presence and absence of DEX (1 µM) for 24 h. As internal standards, BMSCs were homogenized with CHCl₃/CH₃OH/Tris-HCl containing 10 pM of [²H]5-2AG. After that, drying, weighing, and pre-purification of the organic phase containing lipids were performed. The fractions obtained were then analyzed by LC/MS (Shimadzu, Kyoto, Japan). Analyses of 2AG were conducted using the selected mode of ion monitoring for LC/MS. We calculated 2AG levels based on the ratio of their area to the internal deuterated standard signal areas and normalized their amounts per mg of lipid extract to pM.

Animals and chemicals

The Laboratory Animal Center of Soochow University provided male Sprague Dawley (SD) rats (age: 10 weeks, weight: 400 ± 50 g). All rats (5 per cage) were housed in the specific pathogen-free (SPF) animal facility at Suzhou University Experimental Animal Center and were provided with standard feed and sterilized water. The rooms were maintained at constant temperatures and humidity levels, with 12-hour light cycles. As male rats typically have higher bone mass compared to female rats, in order to minimize gender-induced data bias, we exclusively used male rats in all of our experiments. No significant adverse events occurred in our study. As described in our previous study [4], the ONFH model induced by GC was established. Fifty SD rats were randomly assigned to one of five groups (n = 10): (1) Control group (DMSO only); (2) Model group (administration drugs: MP and LPS); (3) Treatment group (administration drugs: MP + LPS + MJN110); (4) AM251 + MJN110 group (administration drugs: MP + LPS + MJN110 + AM251); (5) AM630 + MJN110 group (administration drugs: MP + LPS + MJN110 + AM630). The DMSO, LPS, MP, MJN110, AM251, and AM630 doses utilized were based on those reported in earlier studies [18–23]. Six weeks after the establishment of the model, samples of the femoral head were obtained. All animal experiments were approved by Soochow University’s First Affiliated Hospital’s Ethics Committee.

Micro-CT scans

Using a SkyScan 1176, the femoral heads of the rats were scanned and analyzed (Bruker Micro-CT, Aartselaar, Belgium). Scan settings were as follows: 70 kV, 120 µA, and 200 ms at a voxel resolution of 12 μm in medium mode. For the bone parameter analysis of trabecular bone, the region of interest extends from 0.1 millimeters below the growth plate of the femoral head and is 1.5 mm long. DataViewer was used to acquire the axial section, sagittal section, and coronal section of each specimen. CT Analyzer software was utilized to evaluate the bone volume (BV), bone volume fraction (BV/TV), trabecular thickness (Tb. Th), and trabecular spacing (Tb. Sp).

H&E staining and TRAP staining

The femoral head samples were fixed with paraformaldehyde and decalcified in 10% EDTA. Then, the specimens were embedded and cut into 6-µm-thick sections. After H&E staining or TRAP staining, a laser microscope (Carl Zeiss, Oberkochen, Germany) was used to study the morphological changes by H&E staining and visualize the osteoclast density by TRAP staining. The detailed H&E staining and TRAP staining procedures were according to the previous study [24]. For the TRAP staining experiment, the cells containing purple staining and multiple nuclei were regarded as osteoclasts and manually counted. The number of osteoclasts within the view field was used to evaluate the level of bone resorption.

Immunohistochemistry (IHC) and tissue IF staining

In brief, after antigen retrieval with sodium citrate solution, the sections were stained overnight at 4 °C with RANKL primary antibody. The samples were then treated with the matching secondary antibodies. The positive staining was then developed using a DAB staining kit (Beyotime, China). The portions were then inspected using a laser microscope (Carl Zeiss, Germany). For tissue, IF staining, slices were blocked with BSA for 1 h and incubated with MAGL, OCN, and PPAR primary antibodies for 12 h. The slices were then washed with PBS and incubated for one hour with the matching secondary antibodies. DAPI was used to stain the nucleus. The tissue slices were then observed using a laser microscope (Carl Zeiss, Germany). The number of IHC-positive or IF-positive cells was quantified using Bioquant Osteo 2017.

Masson staining

After dewaxing and rehydrating the sections, they were stained with Weigert’s hematoxylin solution for 10 min. Afterward, they were separated with acidic alcohol and rinsed with ddH₂O. Masson dye was then applied to the sections. Following washing with ddH₂O, the working solution for the Ponceau stain was applied for 10 min. Washing the sections was carried out using a weak acid solution. Afterward, the sections were stained for 2 min with phospholipid acid solution, 1 min with a weak acid working solution, 2 min with aniline blue solution, and 1 min with a weak acid working solution. They were then dehydrated, hyalinized, and mounted using neutral resin.

Statistics

At least three times were conducted for each experiment. SPSS (version 20) was used for statistical analysis. The figures represent the average plus the standard deviation. Comparisons between two groups were performed using unpaired t-tests with Welch’s correction applied to the data. Using ANOVA, among multiple group differences were analyzed. Tukey’s test and Dunnett’s test were used to do the post hoc analysis. The significance level was determined by a P value of 0.05 or a P value of 0.01.

Results

GCs change the fate decision of BMSCs and promote MAGL expression

First, the toxicity of DEX against BMSCs was analyzed with CCK8. We found that when the concentration of DEX was greater than 1.0 µM, it inhibited BMSC proliferation (Figure S1 (A)). Next, we tested whether BMSC differentiation correlated directly with DEX concentration. Western blotting and PCR results showed that the adipogenic-relevant protein and gene (PPARγ, CEBPβ, and LPL) expression levels showed a concentration-dependent increase (Figure S2 (A-G)). In addition, under high concentrations of DEX treatment, a large number of fat droplets were observed in cells (Figure S2 (H-I)). In contrast, the decrease in ALP, RUNX2, and OCN expression levels was accompanied by an increased concentration of DEX (Figure S2 (J-Q)). Meanwhile, the number of ALP-positive cells and calcium nodules decreased significantly (Figure S2 (R-U)). We also examined the levels of proteins and genes associated with osteogenic or adipogenic differentiation in BMSCs treated with 1 µM DEX over various time intervals. The results showed that, with time, the expression of adipogenic differentiation markers rose, whereas the expression of osteogenic differentiation markers decreased (Figure S3 (A-F)). Interestingly, western blotting, PCR, and immunofluorescence staining results also suggested that MAGL expression was strongly associated with DEX intervention (Fig. 1 (A-E)). Next, we conducted an immunofluorescence staining experiment to examine the in vitro results in our ONFH rat model, and we observed more PPARγ- and MAGL-positive cells in model group samples (Fig. 1 (F-I)). Thus, the above results demonstrate that GC can promote BMSC adipogenic differentiation and inhibit osteogenic differentiation. At the same time, it also stimulated MAGL expression.

Fig. 1.

DEX activates MAGL expression in BMSCs. (A-C) Western blot and RT-PCR analyses of MAGL expression levels in BMSCs challenged with DEX for various lengths of time. (D-E) MAGL immunofluorescence staining and quantification of BMSC cultures under normal conditions treated with vehicle or 1 µM DEX (n = 3; mean ± S.D.). (F) IF staining of PPARγ, OCN, and MAGL. (G-I) Average number of IF-positive cells per field. (Control group versus Model group). Data represent the mean ± S.D. (n = 3). All these studies were performed with at least 3 biological replicates

MAGL governs the differentiation of BMSCs in vitro

We analyzed the transcriptomes of shNC- and shMAGL-treated primary SD rat BMSCs to determine if MAGL was involved in the differentiation and development of BMSCs. Figure S4 demonstrates that principal component analysis (PCA) revealed different mRNA expression grouping between the shNC control and MAGL-depleted conditions, which were evaluated in triplicate. We identified 2,376 genes whose expression levels varied. With MAGL depletion, the expressions of 1,168 genes were increased and 1,208 genes were downregulated (Fig. 2 (A)). The genes with upregulated expression contained numerous osteogenic differentiation-related genes, while a large volume of adipogenic differentiation-related genes were downregulated (Fig. 2 (B, C)). More interestingly, according to metascape analysis, adipogenic differentiation was one of the top ten biological processes whose expression was decreased when MAGL was silenced, but the enrichment of osteoblast differentiation genes was significantly enhanced (Fig. 2 (D-E)).

Fig. 2.

MAGL blockade promotes the osteogenic differentiation of BMSCs and inhibits their adipogenic differentiation. (A) Volcano plot: The red dots indicate genes whose expression has significantly increased (1,168), while the blue dots represent genes whose expression has significantly decreased (1,208) and genes with black dots are not significantly differentially expressed. (B-C) Hierarchical clustering analysis of all differentially regulated osteogenic differentiation-related genes and adipogenic differentiation-related genes from the BMSCs transfected with shNC or shMAGL for 2 days. (D-E) Meta-scape analysis was performed on genes with up- and down-regulated expression. Based on log10 FC, we plotted the top Ten up- and down-regulated enriched Gene Ontology terms in our dataset. (F) MAGL expression was determined by RT-PCR. In the MJN110 group, BMSCs were pretreated with the MAGL inhibitor MJN110 (1 µΜ) for 48 h. (G-N) Western blot and RT-PCR analyses of the expression of osteogenic and adipogenic marker in control cells and MJN110 cells. In the MJN110 group, BMSCs were pretreated with the MAGL inhibitor MJN110 (1 µΜ) for 48 h. (O) Oil red O staining was performed on day 14 of adipogenic differentiation. (P) Quantitative analysis of the number of adipocytes in (O). (Q-R) ALP staining and ALP activity detection were performed on day 7 of osteogenic differentiation. (S-T) Alizarin red staining and ARS absorbance detection were performed to indicate mineral deposition on day 14. Data represent the mean ± S.D. (n = 3). These studies were performed with at least 3 biological replicates

Next, to verify the above sequencing results, we inhibited the expression of MAGL in BMSCs with an inhibitor (MJN110), and dimethyl sulfoxide (DMSO) was used as a control. First, we verified that MIN110 is not cytotoxic to BMSC at 10 µΜ dose (Figure S5). As a next step, we differentiated BMSCs into osteogenic and adipogenic cells. During adipogenic development, treatment with MJN110 decreased adipogenic marker gene and protein levels significantly compared to control, as shown by Western blotting and PCR analysis (Fig. 2 (G-I), (M), and Figure S6). Meanwhile, MJN110 also significantly increased the osteogenic marker expression levels in BMSCs during osteogenic development. (Fig. 2 (J-L) and (N), and Figure S6). Increased ALP-positive cells and calcium nodules, as well as decreased fat droplets, further confirmed that inhibition of MAGL significantly suppressed BMSC adipogenic differentiation but promoted osteogenic differentiation (Fig. 2 (O-T)). We next transfected BMSCs with MAGL plasmid (OV-MAGL) or blank plasmid control (OV-NC). As expected, MAGL overexpression in BMSCs yielded opposing effects, as described by a significant repression of osteogenic differentiation but an enhancement of adipogenic differentiation compared with OV-NC cells (Figure S7 (A-G)). Collectively, our results implicated MAGL as a potentially negative regulator of BMSC differentiation fate.

MAGL blockade improves ONFH by reversing the effects of GC on BMSC differentiation

Because DEX upregulates MAGL expression, we next investigated whether MAGL blockade could reverse the effects of DEX on BMSC differentiation. Results from Western blotting and PCR demonstrated that the MJN110-pretreated group had lower adipogenic marker expression (Fig. 3 (A-D) and Figure S8 (A)). Oil Red O staining confirmed that DEX-mediated enhancement of BMSC adipogenic differentiation was suppressed by MJN110 (Fig. 3 (E-F)). We also evaluated the impact of MAGL blockage on the osteogenic differentiation of BMSCs after DEX therapy. The results showed that the osteogenic indicator expression levels were significantly elevated after pretreatment with MJN110. (Fig. 3 (G-J) and Figure S8 (B)). In addition, ALP-positive cells and the number of matrix mineralizations were also increased after MAGL inhibition (Fig. 3 (K-N)). Furthermore, we also tested whether adding MJN110 after 48 h of DEX intervention in BMSCs would have a therapeutic effect. The results showed that even with delayed MAGL blockade, it could still effectively inhibit the differentiation of BMSCs into adipocytes and promote their osteogenic differentiation (Figure S9 (A-H)). Overall, our results demonstrated that MAGL blockade could rescue the effects of GC on BMSC differentiation.

Fig. 3.

MAGL blockade rescues the effect of DEX on the direction of BMSC differentiation. (A-D) Western blot and RT-PCR analyses of the expression of adipogenic marker genes. In the MJN110 group and DEX + MJN110 group, BMSCs were pretreated with MJN110 (1 µΜ) for 48 h. (E) Oil red O staining was performed on day 14 of adipo-genic differentiation. (F) Quantitative analysis of the number of adipocytes in (E). (G-J) Western blot and RT-PCR analyses of the expression of osteogenic marker genes. In the MJN110 group and DEX + MJN110 group, BMSCs were pretreated with MJN110 (1 µΜ) for 48 h. (K-L) ALP staining and ALP activity detection were performed on day 7 of osteogenic differentiation. (M-N) Osteogenic differentiation was confirmed by Alizarin red staining and ARS absorbance detection on day 14. Data represent the mean ± S.D. (n = 3). These studies were performed with at least 3 biological replicates

We further investigated whether early treatment with MJN110 could improve GC-induced ONFH. The process of MJN110 administration in rats is sketched in Fig. 4A. In the early stages of ONFH, a deficiency in the ossification capacity of the growth plate results in trabecular bone loss beneath the articular cartilage of the femoral head, and this region will be populated through the migratory infiltration of chondrocytes. MicroCT imaging results showed that after MJN110 treatment, the central trabecular bone was efficiently repaired and regularly arranged in the femoral head (Fig. 4 (B)). Further analysis of bone parameters confirmed the imaging findings. In the MJN110 group, the values of BV, BV/TV, and Tb.Th. rose significantly relative to the model group, whereas Tb. Sp values decreased markedly (Fig. 4 (C-F)). As shown by H&E staining, Masson staining, and Oil Red O staining, with MJN110 administration, the central trabecular bone was significantly restored, the trabecular bone at the center of the femoral head became wider and denser, and the lipid droplets were fewer (Fig. 4 (G)). Based on the above histological analyses, we confirmed that the incidence of ONFH was much lower in the MJN110 group (2/10) than in the model group (8/10). Moreover, using immunofluorescence staining, we further demonstrated that MAGL inhibition reversed the effect of GC on BMSC differentiation and thus promoted ONFH (Fig. 4 (H-K)).

Fig. 4.

MAGL inhibition improved outcomes of GC-induced ONFH. (A) The timeline of the GC-induced ONFH model and administration of MJN110 in vivo. (B) Images of micro-CT. (C) BV (mm³) (D) BV/TV (%). (E) Tb.Th (mm). (F) Tb.Sp (mm). (G) H&E staining, Masson staining, and Oil Red O staining of bone tissues. (H) IF staining of PPARγ, OCN, and MAGL. The IF-positive cells are marked with white arrows. (I-K) Quantitative analysis of the number of IF-positive cells in (H). Data represent the mean ± S.D. (n = 5). All in vitro studies and in vivo studies were performed with at least 3 biological replicates

The function of MAGL inhibition in regulating BMSC differentiation is mediated partially by CB2 but not CB1

To determine whether the regulatory effect induced by MAGL inhibition was mediated by cannabinoid signaling, we used siRNAs and corresponding small-molecule inhibitors to reduce the expression of CB1 and CB2 (Figure S10 (A-F)). The PCR analyses indicated that CB1 inhibitor treatment or siCB1 transfection enhanced osteogenic differentiation but attenuated adipogenic differentiation when the interventions were individual and exerted additive effects when treated with MJN110 (Fig. 5 (A and C)). In contrast, the decreased levels of adipogenic markers and the increased levels of osteogenic markers in MJN110-treated BMSCs were significantly but not completely reversed by CB2 blockade (Fig. 5 (B and D)). Moreover, we observed a significant reduction in ALP-positive cells and a clear increase in fat droplets when siCB2-transfected cells were treated with MJN110 (Fig. 5 (E-H)). MAGL is known to be the main 2AG glycohydrolase in vivo, and MAGL blockade directly leads to elevated levels of 2AG [25]. We confirmed that MJN110 was efficacious in improving the DEX-induced 2AG level decrease in BMSCs (Figure S11). Since 2AG is the primary ligand for CB2 receptors, we further investigated the effect of 2AG intervention on BMSC differentiation. Interestingly, we also observed that the 2AG intervention reversed the effect of GC on BMSC differentiation, which was similar to the MJN110 treatment (Figure S12 (A-T)). Consistent with the above in vitro results, we found that the CB2 antagonist AM630 partially reversed the MJN110 effects in vivo (Figure S13 (B-J)). Compared with the MJN110 group, reduced subchondral trabecular bone and more lipid droplets were observed in the AM630 group (Figure S13 (G)). Meanwhile, Masson staining and analysis of bone parameters suggested that osteoporosis became more severe in the AM630 group (Figure S13 (G)). Remarkably, a CB1 antagonist (AM251) exerted additive protective effects on the femoral head when administered with MJN110 (Figure S13 (B-J)). Collectively, our data demonstrate that the regulatory effects of MAGL inhibition on BMSCs are partly mediated by the enhancement of endocannabinoid signaling acting on CB2.

Fig. 5.

The osteogenic differentiation promoting effect of MAGL blockade is due in part to CB2, but not CB1 signaling. (A) The effect of the CB1 receptor inhibitor AM251 on ALP, OCN, and CEBPβ expression levels after 48 h of DEX (1 µΜ) treatment. AM251 inhibited CEBPβ expression while promoting ALP and OCN expression in BMSCs and this effect is additive when AM251 and MJN110 are combined. (B) The effect of the CB2 receptor inhibitor AM630 on ALP, OCN, and CEBPβ expression levels after 48 h of DEX treatment. AM630 inhibited ALP and OCN expressions while promoting CEBPβ expression in BMSCs and AM630 weakens the effect of MJN110 on BMSCs. (C) Inhibition of MAGL after 48 h of DEX (1 µΜ) treatment exerted further increases of ALP and OCN in CB1 knockout BMSCs. (D) Knockdown of CB2 attenuated the promoting effect of MJN110 on BMSCs adipogenic differentiation induced by DEX (5 µΜ). (E) Oil red O staining was performed on day 14 of adipogenic differentiation, showing an enhanced adipogenic differentiation in the DEX + MJN110 + siCB2 group compared to the DEX + MJN110 group. (F) Quantitative analysis of the number of adipocytes in (E). (G-H) ALP staining and ALP activity detection were performed on day 7 of osteogenic differentiation, showing an attenuated osteogenic differentiation in the DEX + MJN110 + siCB2 group compared to the DEX + MJN110 group. Data represent mean ± S.D. (n = 3). These studies were performed with at least 3 biological replicates

MAGL blockade regulates the CB2-mediated PI3K/AKT/GSK3β pathway during BMSC differentiation

CB2 plays crucial roles in various physiological and pathological processes by controlling multiple signaling pathways [26–29]. One of these, the PI3K/AKT/GSK3 pathway, has been shown to reduce osteoporosis caused by GC [30]. Based on sequencing results, one of the top twenty KEGG enrichment pathways that shMAGL upregulated was the PI3K/AKT cascades (Fig. 6 (A)). To further comprehend the mechanisms by which MAGL inhibition affects BMSC development, we examined whether MAGL inhibition reverses the effects of GC on BMSC differentiation through the PI3K/AKT/GSK3β pathway. First, by western blot analyses, we demonstrated that p-AKT (S473) and p-GSK3β (Ser9) expression were markedly downregulated in the DEX-treated group (Fig. 6 (B-D)). When MJN110 was added to the conditioned medium, the PI3K/AKT/GSK3β pathway was activated (Fig. 6 (B-D)). We also detected the protein expression levels of p-AKT and p-GSK3β in the bone tissue of the model group and the treatment group, and the results were consistent with the in vitro findings (Figure S14 (A-C)). Furthermore, we found that the recovery effects of MJN110 on the suppression of the PI3K/AKT/GSK3β pathway by DEX were reversed by CB2 inhibitor AM630 (Fig. 6 (E-G)). Thus, the above experimental results suggested that the regulation of the PI3K/AKT/GSK3β pathway by MAGL was mediated via CB2. Next, to determine whether the impact of MAGL inhibition on BMSC differentiation was mediated via the PI3K/AKT/GSK3β signaling pathway, we suppressed PI3K expression in BMSCs using LY294002. LY294002 blocked the PI3K signaling pathway efficiently (Figure S15 (A-E)). Western blot analysis revealed that LY294002 dose-dependently increased the reduced expression of adipogenic markers mediated by MAGL inhibition. Conversely, the expression levels of osteogenic markers showed a decreasing tendency with increasing LY294002 concentration (Fig. 6 (H) and Figure S16). Moreover, we also observed considerably reduced ALP-positive cells, calcium nodules and a considerable increase in fat droplets after LY294002 treatment (Fig. 6 (I-L) and Figure S17 (A-B)). These data demonstrate that activation of the PI3K/AKT/GSKβ signaling pathway is required for MAGL inhibition to control BMSC differentiation in GC.

Fig. 6.

MAGL blockade regulates the CB2-mediated PI3K/AKT/GSK3β pathway during osteogenic differentiation in BMSCs. (A) KEGG enrichment analysis of the upregulated pathways. (B-D) Western blot was performed to analyze the PI3K/AKT/GSK3β signaling pathway-related protein expression levels. BMSCs in DEX group were treated with DEX (1 µΜ) for 7 days; BMSCs in DEX + MJN110 group were pretreated with MJN110 (1 µΜ) for 48 h and then treated for 7 days with DEX (1 µΜ) and MJN110 (1 µΜ) in combination. Data represent mean ± S.D. (n = 3). (E-G) Western blot was performed to analyze the PI3K/AKT/GSK3β signaling pathway-related protein expression levels. BMSCs in DEX + MJN110 group were pretreated with MJN110 (1 µΜ) for 48 h and then treated for 7 days with DEX (1 µΜ) and MJN110 (1 µΜ) in combination; BMSCs in DEX + MJN110 + AM630 group were pretreated with MJN110 (1 µΜ) for 48 h and then treated for 7 days with DEX (1 µΜ), MJN110 (1 µΜ) and AM630 (1 µΜ) in combination. Data represent mean ± S.D. (n = 3). (H) Western blot was performed to analyze the osteogenic and adipogenic-related protein expression levels. BMSCs were pretreated with MJN110 (1 µΜ) for 48 h and then treated for 14 days or 7 days with DEX (1 µΜ), MJN110 (1 µΜ), and various concentrations of LY in combination. (I-J) ALP staining and ALP activity detection were performed on day 7 of osteogenic differentiation. LY attenuated the osteogenic differentiation in a concentration dependent manner. Data represent mean ± S.D. (n = 3). (K) Oil red O staining was performed on day 14 of adipogenic differentiation, LY enhanced the adipogenic differentiation in a concentration dependent manner. (L) Quantitative analysis of the number of adipocytes in (K). Data represent mean ± S.D. (n = 3). These studies were performed at least 3 biological replicates

Blocking MAGL prevents GC-induced bone resorption by regulating osteoclastogenesis

Finally, we tested whether MAGL blockade also improves GC-induced ONFH by suppressing bone resorption. Strikingly, we found that MJN110 treatment decreased osteoclast numbers in femoral heads (Fig. 7 (A-B)). RAW264.7 cells were then incubated in a conditioned medium containing different MJN110 concentrations to examine whether MJN110 could affect osteoclast formation. TRAP staining results indicated that the number of osteoclasts decreased as the MIN110 concentration increased. (Figure S18 (A-B)). RANKL is a key mediator of osteoclast development, and GCs stimulating RANKL expression in osteoblasts are another important pathway to enhance bone resorption [31]. Thus, we additionally measured RANKL and OPG expression levels in BMSCs and bone tissues under different culture conditions. The results showed that DEX markedly reduced OPG expression but elevated RANKL expression. It directly led to an increased RANKL/OPG ratio (Figure S19 (A-E) and Fig. 7 (C-H)). In addition, immunohistochemistry findings also confirmed that RANKL expression was elevated in the femoral heads of the model group. This trend was reversed by MJN110 treatment and was consistent with the TRAP staining results (Figure S19 (A-E) and Fig 7 (C-H)). Together, these data suggest that MAGL blockade could reduce osteoclast formation via direct and indirect impaction and ultimately promote GC-induced bone resorption enhancement.

Fig. 7.

Blocking MAGL prevented osteoclast formation by regulating the RANKL/OPG ratio in BMSCs. (A) TRAP staining of bone tissues. The osteoclasts are marked with black arrows. (B) Quantitative analysis of the number of TRAP-positive cells in (A). (C-F) Western blot analyses of the protein expression levels of RANKL and OPG in bone tissues. (G) IHC staining of RANKL. The IHC-positive cells are marked with black arrows. (H) Quantitative analysis of the number of IHC-positive cells in (G). Data represent the mean ± S.D. (n = 5). All in vitro studies and in vivo studies were performed with at least 3 biological replicates

Discussion

Increasing evidence suggests that the cannabinoid system is a critical factor involved in bone homeostasis [32]. CB2 receptors have been demonstrated to be highly expressed in bone cells, and CB2 receptor activation is favorable for enhancing bone formation [32]. Moreover, several early studies have also shown that CB1 receptors have important impacts on osteoblast proliferation and differentiation [14]. As a significant node of the cannabinoid system, MAGL typically influences CB1 and CB2 receptor activity by regulating 2AG levels [33]. To date, it remains unclear whether MAGL exerts effects on bone homeostasis. In this study, we observed an increase in the expression of MAGL that synchronously followed an increase in the adipogenic-related markers in BMSCs exposed to DEX. We thus consider the possibility of MAGL acting as a key regulator in BMSC differentiation. By RNA sequencing, MAGL was shown to be a negative regulator of androgenic activity, which was an unexpected result. Further experiments also showed that MAGL overexpression can effectively promote adipogenic differentiation while attenuating osteogenic differentiation of BMSCs, which suggested that MAGL blockade may be an effective strategy for restoring bone homeostasis.

Supraphysiological doses of glucocorticoids would lead to a reduction in BMSC lifespan and shift BMSCs from osteoblasts to adipocytes, although they are endowed with powerful anti-inflammatory effects [6, 34, 35]. It has been reported that glucocorticoids prevent BMSC osteogenic differentiation by blocking the Wnt/β-catenin pathway and enhance adipogenic differentiation by activating CEBPα [6, 36]. Our study confirmed that when the DEX concentration reached 100 nM, it significantly enhanced adipogenesis while suppressing the osteogenic differentiation potential of BMSCs. To date, no study has reported that the cannabinoid system can ameliorate GC-induced suppression of osteogenic differentiation or enhancement of adipogenic differentiation in BMSCs. In our GC-induced ONFH model, we observed that MAGL blockade could effectively reverse the direction of BMSC differentiation changes attributable to GC. In particular, in vivo experiments suggested that an MAGL antagonist partially reversed ONFH by promoting bone formation. With further biomedical development, which can restrict the inhibition of MAGL expression in bone, MAGL blockade may provide a new therapeutic option for GC-induced ONFH.

Previous studies have demonstrated that glucocorticoid activates CB1 expression and that CB1 antagonists could promote GC-induced inhibition of osteoblast differentiation [37]. In contrast to the CB1 receptor, CB2 activation enhances BMSC osteogenic differentiation [4]. MAGL blockade is known to increase 2AG levels in vivo, and 2AG is the common ligand of CB1 and CB2 [38]. However, interestingly, we found that MJN110 is considerably more similar to CB2 agonists but not CB1 agonists. MAGL blockade can achieve a better treatment effect when combined with CB1 antagonists. This finding might be because the expression level of CB2 is higher than that of CB1 in BMSCs, while CB1 resides more in sympathetic nerve terminals [39, 40]. This also suggested that MAGL inhibition selectively stimulated CB2 but not CB1 by elevating the 2-AG level, and the combination of an MAGL antagonist plus peripherally restricted CB1 inhibitors may provide a better therapeutic outcome than monotherapy.

As an essential lipid kinase, PI3K is engaged in a variety of cellular processes, including determining cell survival differentiation and regulating attachment structures. When the PI3K/AKT signaling pathway is engaged, GSK3β becomes phosphorylated [41]. The β-catenin destruction complex is disrupted by phosphorylated GSK3β, which causes the accumulation of β-catenin [42]. The nuclear translocation of β-catenin enhances osteogenic differentiation but inhibits adipogenic differentiation [43, 44]. GC suppresses the PI3K/AKT signaling cascade, whereas PI3K/AKT pathway activation can considerably attenuate GC-induced inhibition of osteoblast differentiation [45]. Our study also confirmed that DEX suppressed the PI3K/AKT signaling pathway, and PI3K agonist treatment significantly enhanced the osteogenic potential of BMSCs while suppressing adipogenesis.

Our in vitro experiments showed that MAGL blockade rescued GC-induced PI3K/AKT/GSK3β signaling pathway inhibition, suggesting that MAGL is upstream of the PI3K signaling pathway. Since MJN110 leads to activation of the CB2 receptor, MAGL inhibition likely causes PI3K pathway activation via the CB2 receptor. Several studies have reported that activated CB2 receptors can positively regulate the PI3K signaling pathway [46, 47]. Hence, by stimulating the PI3K signaling pathway, it is not unexpected that MAGL inhibitors can reverse the effects of GC on BMSC differentiation. Furthermore, we observed that LY294002 appears to counterbalance the beneficial impacts of MJN110 on BMSC differentiation. This further illustrated that MAGL blockade rescues the harmful effects of GC on BMSC differentiation by regulating the PI3K signaling pathway.

Evidence indicates that GCs cause osteoporosis by enhancing osteoclast formation [48]. Conversely, another research has shown that inhibiting MAGL might prevent the maturation of osteoclasts [49]. Our data showed that the MAGL inhibitor was indeed able to reduce osteoclast formation in the context of DEX intervention. Notably, we also found that, compared with the model group, MAGL blockade decreased RANKL expression and the RANKL/OPG ratio in BMSCs. Previous experiments have confirmed that CB2 activation facilitates GSK3β phosphorylation and leads to the initiation of the Wnt/β-catenin signaling pathway [4]. The Wnt/β-catenin signaling pathway negatively modulates the expression of RANKL in osteoblasts [50]. This may explain why CB2 has a significant effect on reducing RANKL expression in osteoblast precursor cells. Thus, we concluded that targeting MAGL inhibition can attenuate bone resorption induced by GCs by directly and indirectly suppressing osteoclast generation. However, further investigation is needed to examine whether other signaling pathways are implicated in the regulation of RANKL by MAGL.

Conclusion

In conclusion, our study confirmed that MAGL participates in the regulation of BMSC fate and that MAGL inhibition can effectively reverse the effect of GCs on BMSC differentiation by activating the PI3K/AKT/GSK3β signaling pathway. We also demonstrated that the modulation of the PI3K pathway generated by MAGL inhibition was mediated by CB2. Notably, MAGL blockade not only promotes bone formation but also attenuates GC-induced bone resorption enhancement, which ultimately alleviates GC-induced bone homeostasis imbalance and ONFH (Fig. 8). Therefore, MAGL and its downstream effectors (CB1 and CB2) may provide new opportunities for treating GC-induced ONFH. However, the potential mechanisms of MAGL downregulation still require further exploration.

Fig. 8.

A schematic of MAGL blockade-mediated protection from GC-induced bone homeostasis imbalance. MAGL blockade suppresses GC-induced osteogenic differentiation inhibition through CB2-mediated PI3K/AKT/GSK3β pathway. MAGL block-ade could also convey a femoral head protection by attenuating the enhancement of GCs-induced bone resorption.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

Ning Yang: Data curation, Formal analysis, Investigation, Writing-original draft, Writing-review & editing. Meng Li: Formal analysis, Investigation, Writing-original draft, Writing-review & editing. Xuefeng Li: Investigation, Writing-original draft, Writing-review & editing. Lunan Wu: Formal analysis, Investigation, Writing-review & editing. Wenzhi Wang: Investigation, Writing-review & editing. Yaozeng Xu: Resources, Writing-review & editing. Zhen Wang: Resources, Writing-review & editing. Chen Zhu: Funding acquisition, Writing-review & editing. Dechun Geng: Funding acquisition, Project administration, Writing-review & editing.

Funding

This work is supported by grants from the National Natural Science Foundation of China (Nos. 82472525, 82072425, 82072498, 82272157 and 82272512), the Natural Science Foundation of Anhui Province, Distinguishing Youth Project (2108085J40), Anhui Provincial Department of Education Higher Education Research Program (2022AH010076), the Young Medical Talents of Jiangsu Province (No. QNRC2016751), the Natural Science Foundation of Jiangsu Province (Nos. BK2021650), the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD) Jiangsu Medical Research Project (ZD2022014), the Program of Suzhou Health Commission (GSWS2022002), National and Local Engineering Laboratory of New Functional Polymer Materials (SDGC2205), Jiangsu Orthopedic Medical Devices and Biomaterials Engineering Research Center, Suzhou Key Laboratory of Orthopedics, and the Foundation of National Center for Translational Medicine (Shanghai) SHU Branch (No. SUITM-202403).

Data availability

The data sets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethical approval

All experiments were conducted in the Soochow University Orthopaedic Institute. The study protocol and the use and care of animals were reviewed and approved by the Tab of Soochow University’s First Affiliated Hospital’s Ethics Committee, Suzhou Medical College of Soochow University (202112A0630).

Consent for publication

The authors consent to publishing this work.

Conflict of interest

The authors declare that they have no conflicts of interest for this work.